Air pollution levels have decreased significantly over the previous decades, yet increasing evidence reveals that air pollution remains harmful even at current levels. These levels tend to be higher in densely urbanised zones; in cities. As such, almost everybody is exposed to outdoor air pollution. The health outcomes include the full spectrum from mortality via hospital admissions to sub-clinical effects. Adverse health effects occur both as a result from short term exposure (on the order of days) and from sustained exposure (multiple years). In the BruxAir project, short term effects of air pollution on health were investigated by means of administrative data: the InterMutualistic Agency keeps a record of all prescription-based purchases for residents of Brussels (and the rest of Belgium). The date and quantity of sales were linked to personal characteristics (age, sex, place of residence) and to exposure measurements (daily, 4x4km² resolution). The aggregated data were analysed with time series methodologies. The objectives of the study were to (1) evaluate the use of the purchase of respiratory medication sales as a proxy for the adverse health effects, (2) optimize the statistical methodology, (3) estimate the relative risks and (4) estimate the cost saving potential. Results include (1) purchase is indeed a sensitive indicator, (2) ensembles and compartment methods are promising methods to deal with these classes of problems, (3) significant relative risks are found, in particular for NO2 (a known proxy for traffic-related pollution) and for grasses (the plant pollen that people are most commonly sensitized to) and (4) an annual cost saving potential of €73k per 10% decrease in NO2 concentrations in the Brussels Capital Region, counting only the sales of reimbursable respiratory medications for asthma and/or chronic obstructive pulmonary disease.
